Do DLX3 and CD271 Protect Human Keratinocytes from Squamous Tumor Development?

Well-regulated epidermal homeostasis depends on the function of different classes of factors, such as transcription regulators and receptors. Alterations in this homeostatic balance may lead to the development of cutaneous squamous tumorigenesis. The homeobox transcription factor DLX3 is determinant for a p53-dependent regulation of epidermal differentiation and modulates skin carcinogenesis. The maintenance of skin homeostasis also involves the action of neurotrophins (NTs) and their receptors, Trk and CD271. While Trk receptor overexpression is a hallmark of cancer, there are conflicting data on CD271 expression and function in cutaneous SCC (cSCC). Previous studies have reported NT receptors expression in head and neck SSC (HNSCC). We show that CD271 is expressed at low levels in primary cSCC cells and the number of CD271+ cells correlates with cell cohesion in SCC spheroids. In normal epidermis, CD271 is expressed in proliferative progenitor cells and DLX3 in terminally differentiated keratinocytes. Brain-derived neurotrophic factor (BDNF) and neurotrophin 3 (NT3) increase DLX3 expression. In the absence of a functional BDNF receptor TrkB in keratinocytes, we hypothesize that the BDNF-dependent DLX3 response could be mediated via CD271. Altogether, our results support a putative CD271-DLX3 connection in keratinocytes, which might be crucial to preventing squamous skin cancer.


Introduction
Skin is the largest organ of the body, with a multitude of functions that depend on a well-regulated epidermal homeostasis. This homeostatic balance is dependent on the equilibrium between cells, mainly keratinocytes, and the function of transcription factors, receptors, growth factors, enzymes, and structural molecules [1]. Keratinocyte stem cells reside in the stratum basale and divide, yielding a limited number of early differentiated progenitor cells that continue to proliferate and then commit to differentiation. During the differentiation process, keratinocytes permanently exit the cell cycle, initiate expression of epidermal differentiation markers, and migrate supra-basally towards the surface of the epidermis. Any alteration of this fine balance may cause a pathological condition of the skin. In this context, cutaneous squamous cell carcinoma (cSCC) is the malignant prototype of altered epidermal homeostasis associated with uncontrolled proliferation, aberrant differentiation, and altered apoptosis.
cSCC represents the second most frequent type of skin cancer, comprising about 20% of skin malignancies and showing a rapidly increasing incidence worldwide [2]. cSCC constitutes a spectrum of progressively advanced malignancies, ranging from a precursor actinic keratosis (AK) to squamous cell carcinoma (SCC) in situ, invasive SCC and metastatic SCC. While differentiating, keratinocytes acquire a defined gene expression signature that includes cell cycle regulators and tumor suppressor (HNSCC) and that inhibition of CD271 has profound negative effects on HNSCC's tumor-initiating capacity [22].
We analyzed the expression of CD271 in different samples of normal healthy keratinocytes (NK), primary cSCC, and in a mucosal SCC cell line (i.e., SCC15) ( Figure 1A). Overall, our results show that CD271 is expressed at significantly lower levels in SCC cells from primary tumors compared to normal epidermal keratinocytes (NK). In contrast, CD271 expression is higher in the SCC15 cell line compared to primary tumors. The anti-CD271 antibody recognizes two bands by Western Blotting analysis, suggesting that CD271 might undergo post-translational processing in primary SCC cells ( Figure 1A). This provides the possibility for a cancer-cell specific mechanism that potentially regulates CD271 expression and function.
show that CD271 is expressed at significantly lower levels in SCC cells from primary tumors compared to normal epidermal keratinocytes (NK). In contrast, CD271 expression is higher in the SCC15 cell line compared to primary tumors. The anti-CD271 antibody recognizes two bands by Western Blotting analysis, suggesting that CD271 might undergo post-translational processing in primary SCC cells ( Figure 1A). This provides the possibility for a cancer-cell specific mechanism that potentially regulates CD271 expression and function.
As alternative models for cSCC studies, we evaluated CD271 expression in different SCC lines, such as SCC12B, SCC13 and SCC15 ( Figure 1B). SCC12B and SCC13 were originally obtained from SCC of the facial epidermis. SCC15 was derived from SCC of the tongue [23]. The percentage of CD271 expressing cells was substantially low in all cell lines, ranging from 2.5 to 8%. However, we observed a lower number of CD271 positive cells in cutaneous versus mucosal carcinoma. Furthermore, the highest percentage of CD271 correlates with a highest compactation in SCC spheroids. Indeed, the total area of SCC15 spheroids is significantly smaller when compared to SCC13 and SCC12B spheroid areas ( Figure 1C), which display a less compact shape ( Figure 1D). These observations are in agreement with the reduced cell-to-cell adhesion and increased area of CD271-silenced melanoma spheroids, and more compact shape displayed by melanoma spheroids overexpressing CD271 [24].  As alternative models for cSCC studies, we evaluated CD271 expression in different SCC lines, such as SCC12B, SCC13 and SCC15 ( Figure 1B). SCC12B and SCC13 were originally obtained from SCC of the facial epidermis. SCC15 was derived from SCC of the tongue [23]. The percentage of CD271 expressing cells was substantially low in all cell lines, ranging from 2.5 to 8%. However, we observed a lower number of CD271 positive cells in cutaneous versus mucosal carcinoma. Furthermore, the highest percentage of CD271 correlates with a highest compactation in SCC spheroids. Indeed, the total area of SCC15 spheroids is significantly smaller when compared to SCC13 and SCC12B spheroid areas ( Figure 1C), which display a less compact shape ( Figure 1D). These observations are in agreement with the reduced cell-to-cell adhesion and increased area of CD271-silenced melanoma spheroids, and more compact shape displayed by melanoma spheroids overexpressing CD271 [24].
The analysis of public transcriptomic datasets from Oncomine (www.oncomine.org) determined there are no significant studies reporting cutaneous squamous carcinomas, while several studies evaluate HNSCC tumors versus normal tissue. We analyzed the expression levels of CD271 and Trk (TrkA, TrkB, TrkC) mRNA in the different datasets, such as the Ginos, Peng, and Cromer head-neck study. The analysis determined highly heterogeneous expression in HNSCC (Figure 2A,B). In the Peng head-neck study, CD271 expression is statistically down-regulated in tumors as compared to normal tissue. In addition, low levels of CD271 mRNA correlates with reduced expression of DLX3 and modulation of other tumor-associated markers [7]. For all results, data represent the mean from three independent experiments ± SEM. * for p < 0.05, ** for p < 0.01.
The analysis of public transcriptomic datasets from Oncomine (www.oncomine.org) determined there are no significant studies reporting cutaneous squamous carcinomas, while several studies evaluate HNSCC tumors versus normal tissue. We analyzed the expression levels of CD271 and Trk (TrkA, TrkB, TrkC) mRNA in the different datasets, such as the Ginos, Peng, and Cromer head-neck study. The analysis determined highly heterogeneous expression in HNSCC ( Figure  2A,B). In the Peng head-neck study, CD271 expression is statistically down-regulated in tumors as compared to normal tissue. In addition, low levels of CD271 mRNA correlates with reduced expression of DLX3 and modulation of other tumor-associated markers [7].

DLX3-CD271 Feedback Loop in Human Healthy Keratinocytes
Given the previously demonstrated role of DLX3 and CD271 in epidermal homeostasis [6][7][8][9], we checked for a functional connection between these factors in healthy keratinocytes. We found B A log2 median-centered intensity log2 median-centered intensity log2 median-centered intensity

DLX3-CD271 Feedback Loop in Human Healthy Keratinocytes
Given the previously demonstrated role of DLX3 and CD271 in epidermal homeostasis [6-9], we checked for a functional connection between these factors in healthy keratinocytes. We found that DLX3 expression after calcium or confluence-induced keratinocyte differentiation overlaps with CD271 expression ( Figure 3A). In particular, calcium was added to subconfluent keratinocytes in order to point out the effect of this pro-differentitative stimulus on proliferating and less differentiated keratinocytes. Furthermore, only the CD271+ sorted cells from total healthy normal keratinocytes (NK) express DLX3. These are the most differentiated cells, as evidenced by the expression of involucrin ( Figure 3B).    . For all results, data represent the mean from three independent experiments ± SEM. For all Western blotting, densitometry was performed by ImageJ software, as described in the Materials and Methods section. * for p < 0.05, ** for p < 0.01, or *** for p < 0.001.
NK were separated according to β1-integrin expression as previously described [11], obtaining three keratinocyte subpopulations characterized by high clonogenic potential and stem cell-like properties, intermediate clonogenic potential with early progenitor features (proliferative progeny), and terminal differentiated phenotypes. Indeed, by colony-forming efficiency and long-term assays, plus the analysis of the epidermal differentiation markers, we have demonstrated that these cells represent distinct epidermal compartments [11]. However, given that β1-integrin is not the unique marker for keratinocyte stem cells, we used the term "stem cell-like" instead of "stem cells". CD271 was expressed by the proliferative progenitors, while DLX3 was expressed primarily in the most differentiated cells ( Figure 3C). In addition, in NK, DLX3 expression was modulated by NTs ( Figure 3D). In fact, recombinant nerve growth factor (NGF) promoted DLX3 downregulation, while brain-derived neurotrophic Factor (BDNF) and neurotrophin 3 (NT3) increased DLX3 expression. We have also shown that the apoptotic marker caspase 3 was activated in BDNF-treated cells and that BDNF is able to decrease cell viability of the proliferative progeny cells ( Figure 3E). These data suggest that in absence of a functional BDNF receptor TrkB in NK [25], the BDNF-dependent DLX3 response could be mediated via CD271. These results are in line with the role of NGF in maintaining keratinocyte "stemness" [25]. On the other hand, DLX3 overexpression induced the upregulation of CD271 together with keratinocyte differentiation [7] (Table 1), indicating a potential DLX3-CD271 feedback loop in skin.

Conclusions
Altogether, our results support a putative CD271-DLX3 connection in keratinocytes and foment further studies to identify the signaling events that orchestrate their mutual regulation in the epidermis. DLX3 and CD271 are required for fine tuning the regulation of the epidermal differentiation process. Loss of expression or signaling alteration of either CD271 or DLX3-dependent pathways are critical for the development of epidermal cell dysplasia and, subsequently, squamous skin cancer (Figure 4).

Conclusions
Altogether, our results support a putative CD271-DLX3 connection in keratinocytes and foment further studies to identify the signaling events that orchestrate their mutual regulation in the epidermis. DLX3 and CD271 are required for fine tuning the regulation of the epidermal differentiation process. Loss of expression or signaling alteration of either CD271 or DLX3-dependent pathways are critical for the development of epidermal cell dysplasia and, subsequently, squamous skin cancer (Figure 4).

Isolation of Primary Keratinocytes from cSCC Tissues
Primary keratinocytes from human cSCC were isolated as previously described [26]. Briefly, tumor samples from human cSCC patients were surgically removed and immediately stored in a sterile test tube containing medium and antibiotics. All tumor samples were collected with written informed consent of patients, according to the Declaration of Helsinki after approval of the Modena Medical Ethical Committee (protocol no. 184/10, 7 December 2010). Tumor tissues were washed with PBS without calcium and magnesium, cut into small fragments and digested in DMEM (Dulbecco's . Alteration of CD271-DLX3 connection in keratinocytes. A CD271 and DLX3 feedback loop is required for the maintenance of the epidermal homeostasis through a fine regulation of the epidermal differentiation and cell cycle processes. The absence of their mutual signals, also due to other factors (X), is involved in pathological alterations that are a prelude to cancer development.

Isolation of Primary Keratinocytes from cSCC Tissues
Primary keratinocytes from human cSCC were isolated as previously described [26]. Briefly, tumor samples from human cSCC patients were surgically removed and immediately stored in a sterile test tube containing medium and antibiotics. All tumor samples were collected with written informed consent of patients, according to the Declaration of Helsinki after approval of the Modena Medical Ethical Committee (protocol no. 184/10, 7 December 2010). Tumor tissues were washed with PBS without calcium and magnesium, cut into small fragments and digested in DMEM (Dulbecco's modified eagle medium) containing 200 U/mL type I collagenase, 200 U/mL dispase and 70 U/mL DNase shaking for 2 h at 37 • C. The digested top tissue mixture was then filtered and centrifuged to collect the cells. Total cells were then seeded onto 3T3 feeder layers as previously described [27], and primary and secondary cell cultures were obtained.
For Western blot analysis, collagen IV coated plates were prepared by seeding a human placenta-derived collagen IV solution (100 µg/mL, Sigma-Aldrich, St. Louis, MO, USA). Total cells from cSCC cultures, either at passage 0 or 1, were seeded on collagen IV pre-coated dishes in keratinocyte growth medium (KGM; Lonza, Walkersville, MD, USA) and maintained in culture until pre-confluent conditions.

Isolation of Primary Keratinocytes from Healthy Skin and SCC Cell Line Cultures
Normal human keratinocytes were isolated from healthy skin biopsies obtained from waste materials from a surgical room. Patient consent for experiments was not required because Italian laws consider human tissue left over from surgery as discarded material. Isolated cells were cultured as described by Pincelli and colleagues [28]. Cells were maintained in culture with KGM until pre-confluence conditions and utilized for Western blot analysis.
Stem cell-like, proliferative progeny, and terminally differentiated keratinocyte population enrichment was performed using the previously reported adhesion to type IV collagen method [11], which is based on the expression of β1-integrin. In detail, stem cell-like keratinocytes express the highest levels of β1-integrin when compared to proliferative progeny, while terminally differentiated keratinocytes do not express β1-integrin and do not adhere to collagen IV coated plates. Total human keratinocytes were left to adhere to collagen IV for 5 min in order to isolate the first population. Non-adherent cells from this plate were transferred to a different collagen IV coated plate and allowed to adhere overnight to obtain proliferative progenitors. Non-adherent cells represent the terminally differentiated cells and were removed from the adherent second population plate. All keratinocyte populations were maintained in KGM for 48 h after isolation and used for Western blot analysis.

Multicellular Spheroids Culture
SCC spheroid formation was obtained by liquid overlay method [29]. Briefly, 96-well plates were coated with 100ul/well of 1.5% agarose (Sigma-Aldrich, St. Louis, MO, USA) diluted in DMEM:Ham's F12 (Lonza, Walkersville, MD, USA) and sterilized by 45 min exposure to UVB radiation after agarose polymerization. Cells were seeded at a density of 3000/well and maintained in SCC medium. 24 h after plating, multicellular spheroids were visible with the microscope and photographed. Pictures were analyzed using ImageJ (Wayne Rasband, NIH, Bethesda, MD, USA) software.

Picture Analysis
Pictures of SCC spheroids were analyzed using ImageJ software (Wayne Rasband, NIH, Bethesda, MD, USA) as previously indicated [30]. To calculate the areas, digital images were processed to 300 pixels/inch and converted to 8 bits. Then, the binary images were subjected to a "clean-up" procedure to eliminate artifacts and with the application "analyze particle", the interested area was measured. The software provides a mask for each spheroid, which represents the analyzed particles, as indicated in Figure 1. The total area, which corresponds to the size of the spheroid, was calculated as the number of total pixels. Six spheroids for each cell line were analyzed and the experiment was performed in triplicate.

Cell Sorting
To obtain CD271-positive cells, normal human keratinocytes were incubated in blocking buffer containing DMEM, 10% FBS, 0.1 M sodium azide, and 4% human gamma globulin (Sigma, St. Louis, MO) for 20 min on ice. After staining cells with 0.5% bovine serum albumin (BSA) in PBS with anti-CD271 antibody (1:100 in PBS, Lab Vision Corporation) at room temperature (RT) for 15 min, cells were resuspended with Alexa Fluor anti-mouse antibody (Invitrogen, Paisley, UK) in 0.5% BSA in PBS for 15 min at room temperature. Following this, cell viability solution (7-AAD solution, BD Via-Probe™, Biosciences Pharmingen, San Diego, CA, USA) was added to the pellet of cells for 15 min at room temperature. Finally, culture medium was added to cells. Negative control was obtained by omitting the primary antibody. Data were collected using a FACS Aria III flow cytometer (BD Biosciences) and analyzed on FACS Diva software (BD Biosciences).